Using Linear Mixed Effects in Helicopter Logging Data

Authors: Lyons, Kevin; Xing, Li; Nelson, John D.

Source: Forest Science, Volume 56, Number 4, August 2010 , pp. 356-365(10)

Publisher: Society of American Foresters

Buy & download fulltext article:

Price: $29.50 plus tax (Refund Policy)

Abstract:

This article applies a linear mixed-effects model (LME) to an unpublished helicopter logging productivity data set. The data were clustered in units, and the silviculture treatments varied between units, where units are cutblocks. When the covariance matrix was considered, it was found that the interunit variance was significant and the intraunit variance was heterogeneous. The significant interunit variance results in significant intraunit correlation, and this indicates the need to use LME to analyze these data. When turn time was considered, unit 4, which had the lowest level of group retention, had a conditional predicted value that was significantly less than the marginal predicted value. When turn weight was considered, unit 2, which had small patch cuts, had a conditional predicted value that was significantly less than the marginal predicted value, and unit 4 had a conditional predicted value that was significantly higher than the marginal predicted value. When productivity was considered, the conditional predicted value for unit 4 was significantly greater than the marginal predicted value, and unit 2 was significantly less than the marginal predicted value. In this data set it is interesting to note that productivity is independent of the number of logs in a turn and that turn weight was relatively insensitive to the available explanatory variables.

Keywords: heteroscedasticity; cluster; correlation; productivity; statistical analysis

Document Type: Research article

Publication date: 2010-08-01

More about this publication?